You May Not Need an Immersive Website

A study published in 2022 in the Journal of Marketing measured what augmented reality try-on changes to the sales of an international cosmetics retailer. Its authors summarise the result (opens in a new tab) in two parts. The effect on sales is positive and "the overall impact appears to be small". It is stronger for less popular products and brands. This is to date the most solid measurement of what an immersive visualisation brings to a purchase decision. It describes a conditional effect, not a general one.

An immersive website, in the sense used here, turns the visit into an experience rather than a reading. It relies on a 3D scene, on a story that unfolds as the page scrolls, known as scroll-driven storytelling, on full-screen animation or on sound. Since WebGPU started running on the iPhone (opens in a new tab), the technical cost of these scenes has fallen. That drop says nothing about the value the scene will bring to the project.

Five questions, asked before the quote, separate the projects where immersion pays from those where it costs. Each rests on a study or on public data.

In 2003, Wendy Moe established in the Journal of Consumer Psychology (opens in a new tab) a typology of visits to an online store, built from page-to-page navigation. She distinguishes four, buying, searching, browsing and knowledge-building. Each has its own likelihood of purchase. An immersive website is designed for browsing. The other three visits arrive with a goal.

In 2023, the Nielsen Norman Group tested websites from several industries that practise scrolljacking, that is, the page taking control of scrolling and replacing the visitor's gesture with an animated sequence. In its report (opens in a new tab), "the majority of our study participants were at least mildly disoriented". One participant got lost and had to go back through the page. Another explained that as a prospect he would have been severely agitated and gone elsewhere. The study notes that the phone screen lengthens the sequences and worsens the disorientation.

The same institute recorded as early as 2014 (opens in a new tab) a recurring remark in testing, that an effect was nice the first time "but now it's getting annoying". A patient who returns every month to book with their osteopath, a client looking for the address of their law firm or a visitor checking a price again see the animated introduction more often than anyone. They are also the visitors who are worth the most to the business.

For a service business, the task visit dominates. According to the EDHEC and Ipsos barometer (opens in a new tab) of April 2024, run online with 1,000 people, 84% of French respondents say they have already booked a specialist appointment through a connected health tool. The figure is self-reported and an online panel probably over-represents digital habits. It still describes the reflex with which a patient lands on a practitioner's website. They look for the booking button and expect it on the first screen.

The question is therefore settled by the share of visits with no immediate goal. A brand whose visitors come to discover what it does can offer them an experience. A firm, a shop or a service provider whose visitors come to get something done puts the experience in their way.

The most useful research on this point is not about 3D but about what images lack. The Baymard Institute, which tests e-commerce sites with users, observes in its state of product pages (opens in a new tab) that 42% of users try to work out the size of a product from the photos and that 37% of the sites tested offer none with a scale reference. One participant abandoned a lamp because she could not read its height from the images, without consulting the written dimensions just below. The uncertainty that blocks a purchase is concrete. It concerns a size, a material, a footprint or a combination.

A field study published in 2022 in the Journal of Interactive Marketing, available as an accepted version (opens in a new tab), calls this block an "imagination gap", the distance between what the customer has to picture and what they manage to picture. Its authors show that augmented reality increases actual purchases when it shows products combined rather than alone, because the combination is precisely what the buyer cannot imagine unaided. In one of the experiments, 26% of the customers in the sample went into the shop after seeing the products in augmented reality. The sample is small, 72 people for the purchase analysis. The setting is a café. The mechanism carries over. Immersion pays when it fills an imagination gap the visitor actually has.

Research on instructional animation has said the same thing for twenty years. Mireille Bétrancourt, Julie Bauer-Morrison and Barbara Tversky concluded as early as 2002, in a review of comparative studies (opens in a new tab), that "most of the time, static and animated graphics are equally effective". Two meta-analyses then pinned down where animation wins. The one by Höffler and Leutner (opens in a new tab) in 2007, across 26 studies, measures a modest overall advantage for animation, d = 0.37 on a scale where 0.2 counts as a small effect and 0.8 as a large one. That advantage triples when the content is a physical procedure to learn. The one by Berney and Bétrancourt (opens in a new tab) in 2016, across 61 studies and 7,036 participants, finds an even smaller overall effect, g = 0.226. Animation beats the still image when the content is itself a movement or a process. A firm's practice areas or the price of a service are nothing like a movement. A mechanism, an assembly, a space to walk through or light on a material, on the other hand, are easier to grasp in motion.

The Journal of Marketing study cited at the opening specifies that the effect of augmented reality is stronger for less popular brands and products. Its authors see in it a way to stimulate demand for niche products and to level the playing field for lesser-known brands. The visualisation stands in for a reputation that is missing. An established brand, whose visitor already knows what they will find, has no such gap to fill and gains little from it.

Awareness also works the other way, on first impressions. In 2012, Alexandre Tuch and his colleagues had 119 screenshots of real websites (opens in a new tab) rated after exposures as short as 17 milliseconds. Two variables weighed on the judgement even at that duration. The sites rated most appealing combined low visual complexity and high prototypicality, that is, resemblance to what the visitor expects of a site in that category. The fifty-millisecond judgement and what it means for a generic site are covered in the article on why websites look the same in 2026 (opens in a new tab). For an immersive site, the implication runs the other way. A page that departs from the prototype of its category and adds visual complexity starts with a first-impression handicap. That handicap is recovered when the visitor knows why they are there. It is not recovered when they wanted a phone number.

Some categories have staging in their prototype. Architecture, cars, luxury, destination hotels and creative studios are expected to come with an experience, to the point that a plain page looks like a step back. The article on what converts depending on what you sell (opens in a new tab) shows that luxury deliberately adds friction to the journey. For a local service, the prototype is made of clarity and signals of trust. Staging adds nothing there that the visitor expects.

In September 2026, mobile accounts for 53.05% of page views (opens in a new tab) in France according to StatCounter. The reference device for judging a page is not the one in the studio that designs it. Alex Russell, an engineer specialising in web performance, updates every year a performance budget (opens in a new tab) calculated on the median Android device. For 2026 he picks a Samsung Galaxy A24 4G, sold for $250 at launch against a $353 global average smartphone price, on a 9 Mbps network with 100 ms of latency. For a page to be usable within three seconds on that device, it has about 2 MB in total, including 0.3 MB of JavaScript. If the JavaScript reaches 0.62 MB, the total drops to 1.2 MB.

The median page already exceeds that budget. According to HTTP Archive's Web Almanac 2025 (opens in a new tab), the median mobile home page weighs 2.6 MB, including 632 KB of JavaScript. Only 48% of mobile sites (opens in a new tab) pass all three Core Web Vitals, the speed and stability indicators Google measures on real visits. A 3D scene is added on top of that base. The Three.js library alone weighs 185 KB compressed (opens in a new tab) before the first model and the first texture. An immersive site therefore starts above the median device's budget before it has displayed anything immersive.

That overrun has a measured price. At the end of 2019, Deloitte and Google tracked for thirty days, with no redesign, 37 brand websites (opens in a new tab) in Europe and America and more than 30 million sessions. A 0.1 second gain on mobile is associated with 8.4% more conversion in retail and 21.6% more visitors reaching the form on lead-generation sites, the case of a firm or a service provider. This is an observation rather than an experiment, the study does not prove that speed causes conversion. The order of magnitude is enough to raise the question. Google had measured in 2017 (opens in a new tab) that the probability of a bounce rises by 113% when load time goes from one to seven seconds.

Award-winning sites show that a heavy scene and a fast page can coexist, at the cost of engineering work detailed in the article on the award-winning sites paradox (opens in a new tab). Deferred loading of the scene, a lighter version for modest devices and readable content before the first 3D frame all have a line in the quote. The device question does not rule immersion out. It sets its real cost.

Motion also excludes part of the audience. WCAG criterion 2.3.3 (opens in a new tab) requires that any animation triggered by an interaction can be turned off and lists parallax scrolling among non-essential animations, because it triggers nausea and migraines in people with vestibular disorders. The Nielsen Norman Group recalls (opens in a new tab) that Apple added the "Reduce Motion" setting in iOS 7 for that reason. The technical solutions exist and are detailed in the article on accessible interactions (opens in a new tab). They add a motion-free version to design and maintain.

It is after launch that an immersive site costs the most. On a conventional site, the text, prices, team and services live in a content management tool and change in a few minutes. On an immersive site, the content is tied to scenes. Changing a service means re-timing an animation, replacing a team member calls for a new shoot or a new 3D object and adding an offer means reopening the story. Every update goes through a developer or a motion designer.

The scale of that work is documented, with one reservation. Framer, the publisher of a website builder, surveyed more than 1,900 web professionals for its State of Sites 2026 (opens in a new tab). 53% of website work consists of edits and fixes. The sample is a vendor's and the survey serves its interests. The figure matches what every team observes, maintenance takes more time than creation. A site where every change requires a specialist multiplies the cost of that half of the work.

Expected lifespan matters too. The agency Orbit Media looked up in the Internet Archive the interval between two redesigns for 50 major brand websites (opens in a new tab), two years and one month on average, against six years and four months for its own clients. An immersive experience built for a launch is amortised over the life of that launch. A foundation built to last is amortised over six years. The two horizons call for different budgets and different deliverables.

The most robust form keeps the two apart. The foundation site, brochure or business, carries the content and changes at the pace of the activity. Immersion lives on a layer added for a set period then removed, a launch page, a product page or a campaign. When the campaign ends, the immersive page is taken down without touching the rest of the site.

The five questions and what they indicate

The first two questions weigh more than the other three, because they describe what the visitor expects. The last three describe what immersion will cost and that cost is negotiated within the project.

An immersive website and a singular website are two different things. Singularity, the one the article on website sameness describes as an asset, lives in decisions of expression, the typography, the colour, the rhythm of a two-hundred-millisecond transition or the way a button responds. None of those decisions requires a 3D scene or a scroll-driven story. An osteopath's brochure site can carry a recognisable signature and remain a page that loads in one second on a Galaxy A24.

WebGPU and the libraries built on it have lowered the cost of building an immersive scene. The five questions bear on whether to build one at all. When the visitor has an imagination gap to fill, when the brand has a reputation to build and when the experience lives on a page made for it, immersion is justified by the same data that advises against it elsewhere. In the other cases, the budget it would have absorbed serves the content, the speed and the interactions the visitor meets on every visit better.